The T4 zero is still nearly 1000uS when it should be 318uS. It's just over 1000uS in the original circuit. T5=72uS which is not bad.
This EQ is hopeless, and changing just one component is never going to fix it.
Interesting if the same level of fuss about RIAA accuracy is taken on during the cutting of the press masters. And if that were off, how would you know?
I recall having an early CD that was mastered to the sound of the LP. Someone was told make it sound like the LP. So they played it, listened to it and made it sound that way. Trouble was, they used the wrong cartridge loading capacitance (or none...) and the CD sounded like it.
One the consumer side, we can just do the best we can and assume everything upstream was done with all the I's dotted and T's crossed.
I recall having an early CD that was mastered to the sound of the LP. Someone was told make it sound like the LP. So they played it, listened to it and made it sound that way. Trouble was, they used the wrong cartridge loading capacitance (or none...) and the CD sounded like it.
One the consumer side, we can just do the best we can and assume everything upstream was done with all the I's dotted and T's crossed.
Well you can be sure that the RIAA in the cutter was accurate, but each company had a 'mastering engineer' who could do just about anything to achieve a company sound, centre the bass, fit the tracks onto the record, ...
I think that a lot of records were mastered to sound good on average (i.e. rolled-off) cartridges. That creates a problem for good cartridges.
The good records that I have sound quite close to the CDs of the same material. It can be hard to tell which is playing in an A/B comparison. My preamp is accurate to within +/-0.2dB.
Ed
The good records that I have sound quite close to the CDs of the same material. It can be hard to tell which is playing in an A/B comparison. My preamp is accurate to within +/-0.2dB.
Ed
Well now, some interesting (and also expected) comments so far confirmed my also expected thoughts when I posted this thread.
What's interesting to me is that this somewhat typical RIAA preamp design was used for decades in just about every consumer product brand.... and even in broadcast equipment..
And yes, to those criticizing it's accuracy and inherent flaws, yet, it lasted for all those years with nary a complaint, before the magic of modern integrated circuitry and digital processing came about.
I did some listening tests with it, and compared the results with a modern IC design that surely wouldn't be shot down by critics with golden ears.
That IC driven preamp has a +/- 0.5% deviation to the 'perfect' RIAA curves.
And ya know what?
I really can't complain about the lowly 3-transistor version posted - any differences are so trivial that only an obsessed nitpicker with a loaded up pile of test equipment would notice.
Certain songs/tracks that I'm well used to, were a pleasent joy to hear again.
What's interesting to me is that this somewhat typical RIAA preamp design was used for decades in just about every consumer product brand.... and even in broadcast equipment..
And yes, to those criticizing it's accuracy and inherent flaws, yet, it lasted for all those years with nary a complaint, before the magic of modern integrated circuitry and digital processing came about.
I did some listening tests with it, and compared the results with a modern IC design that surely wouldn't be shot down by critics with golden ears.
That IC driven preamp has a +/- 0.5% deviation to the 'perfect' RIAA curves.
And ya know what?
I really can't complain about the lowly 3-transistor version posted - any differences are so trivial that only an obsessed nitpicker with a loaded up pile of test equipment would notice.
Certain songs/tracks that I'm well used to, were a pleasent joy to hear again.
I don't know what this one is from, but in general I wouldn't say that. Lipschitz was scathing about commercial RIAA accuracy in 1979, as some reviewers already had been, and plenty of people had been getting it right for decades before that.
This one needs more gain, e.g. via a bootstrap, and new EQ by someone who knows how to do it.
At the local radio station here, they used to have turntables with moving-magnet cartridges connected via much too long cables to the mixing console. Those cables (of the type Klotz unbalanced patch cable) had some sort of carbon shielding as well as the normal shielding, but whoever installed them had connected the ends of the carbon shield to the inner conductor, such that it worked as a shunt resistor of about 10 kohm.
It must have completely messed up the treble response and caused extra thermal noise, but no-one complained about it. They just turned up the treble a bit until it sounded about right.
It must have completely messed up the treble response and caused extra thermal noise, but no-one complained about it. They just turned up the treble a bit until it sounded about right.
I recognize the expressions, and LTspice "Help" says a behavioural voltage source can implement the Laplace transform, although in the voltage controlled voltage source there's a note that the response must end falling with frequency, which the standard RIAA expression doesn't. Have you used this function successfully, perhaps with an asc. file as an example to show how?
On the general point of RIAA accuracy, we can't assume the lathe was right. The Standard for RIAA is defined by assuming a perfect velocity transducer on replay equalised by the three time constants and states that the recorded information should conform to that. In other words, a replay transducer would replay a cut record and equalisation to the cutting head would be tweaked to give a flat response. The cutting head was nothing like linear, so it wasn't simply a case of applying inverse time constants when recording. So, upshot is that recorded accuracy was probably also poor. Especially if that awful two-transistor feedback design was used as part of the reference. Or its direct predecessor using a pentode. Both were inaccurate because there was insufficient gain for equalisation to be defined by feedback without the (variable) open-loop gain being taken into account. And that's even before we get to noting that most people applied the time constants incorrectly so that even passive networks were likely to be wrong. Lipshitz expands on that in his paper, where I think he only found one published expample where the designer got it right. Or that the loading impedance on the cartridge probably didn't conform to what the cartridge manufacturer expected. It's a miracle it worked at all!
I thought, well, WoT might not be keen, but it's something I'd been meaning to have a prod around with for some time, so why not now? It's early days, I'm still staring at the UI and going through options. But it'll be interesting I'm sure. The look & feel of the UI was a bit of a rave from the grave, first developed somewhen in the mid 90's I'd guess, when we developers were all using the MFC/MDI interface of the time 🙂. Even the icons hark back to those days. Still, if it aint broke and developers are thin on the ground, why change it and make work you don't need when there's useful functions to add?
Looking onto the simulation, at a first glance it looks hoorible - but:
The frequency axis starts at 1 HZ and the +4dB peak is at 15 Hz. Between 50 Hz and 20 kHz, the deviation is just +/- 1 dB. No wonder that the listening experience does not give huge differences to a circuit with ruler flat response.
Depending on the amp, where this preamp was part of, with its many coupling capacitors following this preamp, it can be expected that this peak is gone when looking on the overall response. As mentioned, the coupling at the output was a 1uF cap into 47k already forming a HP with 3.3 Hz corner frequency already reducing the peak by 1 dB.
The frequency axis starts at 1 HZ and the +4dB peak is at 15 Hz. Between 50 Hz and 20 kHz, the deviation is just +/- 1 dB. No wonder that the listening experience does not give huge differences to a circuit with ruler flat response.
Depending on the amp, where this preamp was part of, with its many coupling capacitors following this preamp, it can be expected that this peak is gone when looking on the overall response. As mentioned, the coupling at the output was a 1uF cap into 47k already forming a HP with 3.3 Hz corner frequency already reducing the peak by 1 dB.
You're describing somebody named Lipschitz, and "some reviewers", likely a handful of them, and also likely from the internet and/or audio magazines I would assume.
Now put them all into a room.... I'll bet that I've thrown parties and gatherings at my home that would top that headcount.
Let's, for a minute, assume a crowd of 20 reviewers, all fussing, 'scathing' as you say, over the accuracy of a given phono preamp.
Now, let's compare that with the millions (or even billions) of 'common people', the 'end-users', who've purchased audio/phono equipment over time, and who've thoroughly enjoyed what they've purchased... basically satisfied customers.
Customers who've never even seen an O'scope or an RIAA curve, but just want to hear the pleasent sound of their favorite artists on records that they've purchased.
And those hoardes of common people are within a plethora of various living rooms, with different acoustic traits.
And their RIAA commercially-designed equipment is basically the common 2 or 3 transistor type as I posted, and a major player in the game.
And also, let's include their FM stereo broadcast receivers in the mix... we certainly don't want to ignore what broadcast has had to offer their listeners.
Don't you find it somewhat odd that a handful of critics moan and groan over something that literally hoardes of others don't find issue with?
"somebody named Lipschitz" was a so called "expert", a rare species among "hoardes of common people".
https://en.wikipedia.org/wiki/Expert
https://en.wikipedia.org/wiki/Expert
Millions of people eat at McDonalds and don’t complain. It’s not wrong to want something better.
Additional to the suggestion by Rongon (reduce the 560k to 270k => 274k) I made some minor changes to the output cap C2 (now 47uF), the load R5 (now 4.7k). Changes in the C3/4-R6/7 network and including the model of Mooly (RIAA source) the deviations boild down to less than 0.5dB from 10Hz to 20kHz. Further notes: The Miller cap for Q2 was missing in the Spice models; I tuned it to 110pF (originally 100pF). The C6 between the base and emiter of Q1 had no effect at all in my simulations, I removed it.
So in theory, this design seems not so bad, but in reality spread in capacitor values and not-included parasitic terms will spoil a lot and the distortion performance mentioned by Baldin above would require further mods or a more refined design.
So in theory, this design seems not so bad, but in reality spread in capacitor values and not-included parasitic terms will spoil a lot and the distortion performance mentioned by Baldin above would require further mods or a more refined design.
Last edited:
👍 well spotted.
@wiseoldtech If you like simple discrete circuitry have a look at the Radford ZD22 preamp. The line stages are impressive subjectively.
Not a good scan of the phono stage I'm afraid. The right hand stage is just a tape buffer.
Yes but open loop high transconductance JFETs sandwiching passive EQ. Reminiscent of classic tube phonos. Have also tried BJTs with loop feedback EQ but not as good when very few. Better when more complex developed to discreet op-amp level.
Regarding replay Riaa curve don't know if it's about the conformity or the transient behavior but each time I tweaked it a wee bit the records changed tone.